cyberagent-mistral-nemo-webnovels

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>

<details><summary>See axolotl config</summary>

axolotl version: 0.4.1

base_model: cyberagent/Mistral-Nemo-Japanese-Instruct-2408
tokenizer_type: AutoTokenizer


load_in_8bit: false
load_in_4bit: false
strict: false


chat_template: chatml
datasets:
  - path: falche/paradox_test_set_200k_sharegpt
    type: sharegpt
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./outputs/mistral-nemo-webnovels

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

use_wandb: true
wandb_project: mistral-nemo-webnovels
wandb_entity: augmxnt
wandb_name: mi300x-cyberagent_mistral_nemo_webnovels-fft-dsz3

gradient_accumulation_steps: 1
micro_batch_size: 8
num_epochs: 3
optimizer: paged_adamw_8bit
lr_scheduler: linear
learning_rate: 8e-6

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 100
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed: axolotl/deepspeed_configs/zero3_bf16.json
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

</details><br>

outputs/mistral-nemo-webnovels

This model is a fine-tuned version of cyberagent/Mistral-Nemo-Japanese-Instruct-2408 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.6891

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 8e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 64
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 3

Training results

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+rocm6.2
• Datasets 3.0.1
• Tokenizers 0.20.1

Training Infra

Compute sponsored by []HotAisle](https://huggingface.co/hotaisle) on an 8 x MI300X node. See the WandB Run Logs for additional details.